GAE6: Pea Aphids: multiple comparisons

Introduction:

Pea aphids, Acyrthosiphon pisum, can be red or green. Red aphids make carotenoids with genes thought to have jumped from a fungus into the aphid genome some time during recent evolutionary history. Aphid body color is thought to be genetically controlled by a single locus, with red being dominant over green. Ladybird beetles tend to consume red aphids on green plants, and parasitoid wasps preferentially attack green aphids. The predation and parasitism pressures appear to maintain the color variation in natural aphid populations.

While screening pea aphid strains from natural populations collected in France, Tsuchida et al. (2010) found several strains of green aphids producing red nymphs. As the nymphs grew, their body color changed from reddish to greenish, and the adults became green. In a survey of endosymbiotic microbiota in these aphid strains they found a previously unrecognized aphid endosymbiont of the genus Rickettsiella. They hypothesised is that the color change was associated with the Rickettsiella infection.

Methods:

To test this, Tsuchida et al. (2010) injected the bacterium into genetically identical, bacteria-free red aphids (Figure 1). Some aphids did not become infected, resulting in three groups of aphids: a planned control that were not injected, an unplanned but useful group that were injected but did not develop an infection, providing a sham control, and a group that were injected and did become infected.

Figure 1: Example photos of Pea aphids used in the experiment.

To measure color, the researchers used Adobe Photoshop CS (version 8.01, Adobe Inc., San Jose, CA, USA) to measure the hue of 10 pixels from color photos of the aphids. The average hue angle ¹ was then recorded for each aphid.

Statistical Methods:

I tested for differences among the un-injected control (Control), the injected but uninfected aphids (Uninfected), and the injected and infected aphids (Infected) by analysis of variance (ANOVA) using the lm() function from R (R Core Team 2024) and the Anova() function from the car package (Fox and Weisberg 2019). All data manipulation and graphical results were computed using packages provided by the tidyverse metapackage (Wickham et al. 2019) and all tabular results were produced using the kbl() function from the kableExtra package (Zhu 2024).

I next constructed independent orthogonal contrasts for our groups. Because the experiment had three degrees of freedom, I could construct two contrasts. I used the emmeans (Lenth 2024) package to run our planned hypothesis tests.

Finally, I ran a Tukey test in order to protect the experimental error rate. This test allowed me to construct simultaneous confidence intervals for my comparisons.

Results:

Infected pea aphids exhibited distinctly larger hue angles and hence greener colors (Table 1, Figure 2). This result supports the researchers’ hypothesis that infection may lead to color change.

Our hypothesis tests showed that there was a statistically significant difference in color change between the uninfected and infected aphids (P < 0.001). However, there was no significant difference in coloration between the control treatment and the injected but uninfected treatment (P = .324) (Table 2).

Once the Tukey correction was applied, the p value for differences between the control and infected groups was significant, as was the p value for differences between uninfected and infected groups (P < .05). However, the difference between the control and injected but uninfected groups was not significant (P = .5801) (Table 3).

Table 1:

Summary of the analysis of variance for differences in hue angle for pea aphids (Acyrthosiphon pisum) that were infected with Rickettsiella compared to uninfected and control aphids.

	SS	df	F	P
Treatment	2379.3	2	65.6	< 0.001
Residuals	525.9	29

Figure 2: Color change as measured by hue angle for pea aphids (Acyrthosiphon pisum) that were infected with Rickettsiella compared to uninfected and control aphids.

Table 2: Planned hypothesis tests

Contrast	Estimate	SE	df	t	P
uninfected vs infected	36.439	3.181	29	11.454	<0.001
Control vs Sham	-1.886	1.878	29	-1.004	0.324

Table 3:

All pair-wise comparisons of means with the Tukey correction to protect the experimentwise error rate at .05.

contrast	estimate	SE	df	t.ratio	p.value
Control - Uninfected	1.8861	1.8779	29	1.0044	0.5801
Control - Infected	-17.2765	1.7777	29	-9.7187	0.0000
Uninfected - Infected	-19.1626	1.9141	29	-10.0112	0.0000

References:

Fox, J., and S. Weisberg. 2019. An R companion to applied regression. Third. Sage, Thousand Oaks CA.

Lenth, R. V. 2024. Emmeans: Estimated marginal means, aka least-squares means.

R Core Team. 2024. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.

Tsuchida, T., R. Koga, M. Horikawa, T. Tsunoda, T. Maoka, S. Matsumoto, J.-C. Simon, and T. Fukatsu. 2010. Symbiotic Bacterium Modifies Aphid Body Color. Science 330:1102–1104.

Wickham, H., M. Averick, J. Bryan, W. Chang, L. D. McGowan, R. François, G. Grolemund, A. Hayes, L. Henry, J. Hester, M. Kuhn, T. L. Pedersen, E. Miller, S. M. Bache, K. Müller, J. Ooms, D. Robinson, D. P. Seidel, V. Spinu, K. Takahashi, D. Vaughan, C. Wilke, K. Woo, and H. Yutani. 2019. Welcome to the tidyverse. Journal of Open Source Software 4:1686.

Zhu, H. 2024. kableExtra: Construct complex table with ’kable’ and pipe syntax.

Footnotes

1. Hue can typically be represented quantitatively by a single number, often corresponding to an angular position around a central or neutral point or axis on a colorspace coordinate diagram .